Quote:
Originally Posted by Makiabel Moughtarib,
tell us more, what do you think of Abdo Saad & Feghali Polls?! |
Hi Makiabel,
I don't really know the difference between those from Abdo Saad and those from Feghali; although i admit that there is more media attention to the 'Abdo Saad' brand name. He seems to have a better 'market reputation'.
I found his analysis during the 05' elections very good and if you recall very close to the real results. Especially the small differences in Baabda/aley for example; which very few people expected. He also predicted a big change in the mood of the 'North' before everyone else picking up on it and start nagging about Hariri's visit there and the vote/sheikhs buying and sectarianism
But, when any normal reader is checking out poll results, fish for these things, and if you do find them, then the poll is badly done:
1. Do you feel that some of the questions are 'leading'? example: This poll didn't have leading questions; it just had one. For example some questions i saw before were like: 'do you consider GMA 'za3eem awhad'... that is very leading. The chance of getting an accurate answer out of it is slim. The word 'za3eem' has many connotations to it and then the word 'awhad' is also another problem. The question is flawed from the beginning. Thats not the right way to ask about popularity or leadership of a political figure.
2. Do you feel there was no real 'choice' given to the respondent? example: In one Beirut poll, one of the choices was 'man yakhtaraho Michel Aoun'; that is GREAT! you have given the Beiruti respondent, every choice possible...real people (a list of names), and the 'side' if you don't have a name in mind or the list of candidates are not clear (too early). Other polls don't have that; they sometimes completely miss some political parties; that makes the accuracy of the scores lower. For example if you are 'left leaning/leftist' and the pollster didn't give you any choice between LCP and Democratic left and just gave you a list of sectarian parties and the FPM and SSNP; you're most likely to throw in your answer to FPM or SSNP (arguably the more secular); that boosts the FPM/SSNP score higher than normal and mis attributes an LCPer to another party
3. Do you feel that some of the 'analysis is over done? Example: this poll has this all over it, take this for example on talking about the older age group: لرغبة منهم في تجديد الولاء للحزب والعائلة، ; nothing in the data suggests that...and the pollster goes on with more of 'over-analysis'
4. Do you feel the questions are badly written? example: The basic Q was 'yakhtor bi balak'; in English that is 'Top of Mind'; Top of Mind is a horrible way to measure political figures. Ariel is top of mind at around 60%, but more people go out and buy Persil; why? Because Persil has more ads on Tv or its price is lower....in Politics: my Top of Mind is General Aoun or Saad el Hariri but my 'vote' could go to Michel el Murr who installed a phone for us a few years back! So the whole poll is not helpful.
These would be the basic four that anyone can answer; then there are the more
technical ones:
1. Phone interviewing vs. Face to Face: always trust Phone more
2. The bigger the sample size the better (look at polls that are of 500, or 800 people at least)
--For example: the nashra poll makes a big conclusion of how older age groups prefer Sami Gemayel to younger age groups; that is totally unfounded; the whole sample who answered that Q is just 90 people...you can't make such a conclusion; the margin of error is like +/- 20!
Some badly done polls are just because the pollsters want to go to 'field' first and get out data first so they go in with not so well thought through questionnaires. Or don't have enough money to talk to many people so they make conclusions out of a 'bunch' instead of a good representative sample.
Hope it helps.